Progression of a Muscular Dystrophy Due to a Genetic Defect in Membrane Synthesis is Driven by Large Changes in Neutral Lipid Metabolism


 CHKB encodes one of two mammalian choline kinase enzymes that catalyze the first step in the synthesis of the major membrane phospholipid, phosphatidylcholine (PC). In humans, inactivation of the CHKB gene causes a recessive form of a rostral-to-caudal congenital muscular dystrophy. Using Chkb knockout mice, we reveal that at no stage of the disease is PC level significantly altered. Instead, at early stages of the disease the level of mitochondrial specific lipids acylcarnitine (AcCa) and cardiolipin (CL) increase 15-fold and 10-fold, respectively. Importantly, these changes are only observed in affected muscle and contribute to the decrease in the skeletal muscle functional output in these mice. As the disease progresses, AcCa and CL levels normalize and there is a 12-fold increase in the neutral storage lipid triacylgycerol and a 3-fold increase in its upstream lipid diacylglycerol. Our findings indicate that the major changes in lipid metabolism upon loss of function of Chkb is not a change in PC level, but instead is an initial inability to utilize fatty acids for energy resulting in shunting of fatty acids into triacyglycerol.


Introduction
Phosphatidylcholine (PC) is the major phospholipid present in mammalian cells, comprising approximately 50% of phospholipid mass. Choline kinase catalyzes the phosphorylation of choline to phosphocholine and is the first enzymatic step in the synthesis of PC 1 . There are two genes that encode human choline kinase enzymes, CHKA and CHKB. Monomeric choline kinase proteins combine to form homo-or heterodimeric active forms 2 . CHKA and CHKB proteins share similar structures and enzyme activity but display some distinct molecular structural domains and differential tissue expression patterns. Knock-out of the murine Chka gene leads to embryonic lethality 3 .
Chkb deficient (Chkb -/-) mice are viable, but noticeably smaller than their wild type counterparts, and show severe bowing of the ulna and radius at birth. By 2-3 months of age Chkb -/mice lose hindlimb motor control, while the forelimbs are spared 4,5 .
Inactivation of the Chkb gene in mice would be predicted to decrease PC level, however, reports indicate no, or a very modest, decrease in PC level in Chkb -/mice, and this decrease is similar in both forelimb and hindlimb muscle 6,7 . The very small decrease in PC mass, and the fact that there is no rostral-to-caudal change in PC, suggest a poor correlation of the anticipated biochemical defects and observed rostralto-caudal phenotype of this muscular dystrophy 5 . It is unclear how a defect in a gene required for the synthesis the major phospholipid in mammalian cells causes a muscular dystrophy, especially in light of the fact that global inactivation of the CHKB/Chkb gene (human or mouse) does not affect the level of the product of its biochemical pathway, PC.
Muscular dystrophies have been mapped to at least 30 different causal genes 15 . The most common types of muscular dystrophy result from mutations in genes coding for members of protein complexes which act as linkers between the cytoskeleton of the muscle cell and the extracellular matrix that provides mechanical support to the plasma membrane during myofiber contraction 16,17 . Muscular dystrophies result in fibrofatty replacement of muscle tissue, progressive muscle weakness, functional disability and often early death 18 19,20 .
Skeletal muscle accounts for 20-30% of whole body basal metabolic rate 21 . Fatty acid oxidation is the major source of ATP for skeletal muscle during the resting state 22 .
Fatty acids can be synthesized de novo by cells or can be obtained extracellularly, with the bulk of lipids delivered to cells through the circulation via serum albumin or lipoprotein/lipoprotein receptors. For fatty acids to be metabolized they are first activated by esterification to fatty acyl-CoA. Subsequently, they have divergent fates depending on the metabolic status of the cells (Fig. 1). The three major fates of fatty acids are 1. conversion to fatty acyl carnitine for subsequent mitochondrial β-oxidation to provide energy, 2. the synthesis of neutral lipid species for storage as triacylglycerol (TG) rich cytoplasmic lipid droplets, or 3. metabolism into phospholipids, such as PC, to maintain membrane integrity. Fatty acids can also directly bind peroxisome proliferatoractivated receptors (PPARs), key players in the regulation of lipid metabolism by altering the expression of genes required for the conversion of fatty acids to fatty acyl-CoA for phospholipid and TG synthesis, and for fatty acid activation to acylcarnitine (AcCa) for entry into mitochondria and subsequent fatty acid β-oxidation 23 .
In the present study, we use mouse and cell models to investigate the temporal changes in lipid metabolism in the absence of the Chkb gene. Results demonstrate that PC level remains essentially unchanged. Instead, this genetic defect in PC synthesis drives large fluctuations in mitochondrial lipid metabolism with an inability to use fatty acids for mitochondrial β-oxidation resulting in a temporal shunting of fatty acids into TG and their storage as lipid droplets. These changes were specific to affected muscle.
This study provides insight into the surprising biochemical phenotype whereby a genetic block in a lipid metabolic pathway does not directly affect the product of its pathway, and instead alters tangential pathways in a manner that explains the rostral-to-caudal gradient of a genetic disease.

Choline kinase deficient mice display hallmark muscular dystrophy phenotypes
To address the extent that mice lacking Chkb function display gross muscular dystrophy phenotypes, we tested muscle function in Chkb +/+ , Chkb +/and Chkb -/mice from 6 weeks to 20 weeks of age using a grip strength assay and a total distance run to exhaustion test. Body weight was also recorded each week at similar times over the duration of the phenotyping experiments. Body weight of the Chkb +/+ and Chkb +/mice showed no difference between groups (Fig. 2a). The Chkb -/mice weighed significantly less than their wild type counterparts at all time points. The average body weight of 6 Chkb -/mice was 33% to 42% less than that of Chkb +/+ mice at week 6 and week 20, respectively.
Forelimb grip strength measurements were performed at three different timepoints and normalized to body weight. The Chkb -/mice had significantly lower (less than half) the normalized forelimb strength than wild type mice at all three timepoints (week 6, 12 and 18) (Fig. 2b). Another measure of neuromuscular function is the resistance to treadmill running, evaluated as the total distance that each mouse is able to run until exhaustion. The test was performed in all groups at three timepoints (Week 7, 13 and 19). The total distance covered by the wild type mice before exhaustion was similar at all 3 time points (Fig. 2c). There was no significant difference between Chkb +/+ and Chkb +/groups, these mice maintained the ability to cover the same total distance before exhaustion (week 7 vs. week 19; non-significant). At week 7, the Chkb -/mice showed a basal level of total distance run that was 50% that of the wild type or Chkb +/mice. Moreover, the Chkb -/mice showed a decline in running performance from week 7 to week 19, with an almost complete inability to run observed by week 19. Gross measurements of neuromuscular strength in whole mice demonstrate that mice heterozygous for Chkb gene display similar phenotypes to wild type mice. Notably, mice lacking both copies of the Chkb gene display a significant decrease in overt neuromuscular phenotypes.
The level of circulating creatinine kinase (CK), a biomarker of sarcolemmal injury, was determined in Chkb +/+ , Chkb +/-, and Chkb -/mice. No significant change in the serum level of CK was observed in Chkb +/heterozygous mice when compared to the wild type. CK activity was 2.5-fold higher in Chkb -/null mice than that of wild type mice ( Fig. 2d).
To determine if the decreased neuromuscular phenotypes observed in the Chkb -/mice were due to a direct effect on muscle itself, maximal specific force generated by freshly isolated extensor digitorum longus (EDL) muscle from the hindlimb of Chkb +/+ , Chkb +/-, and Chkb -/mice at week 20 was determined. EDL muscle fatigue was measured with 60 isometric contractions for 300 ms each, once every 5 sec, at 250 Hz.
There was no significant difference between wild type and heterozygous Chkb mice in regard to specific force decrease during fatigue and specific force generation, (Fig. 2e, f). Chkb -/mice displayed a specific EDL force that was 10% that of Chkb +/+ or Chkb +/mice. In addition, Chkb -/mice were at maximally fatigued levels, that is those observed in Chkb +/+ or Chkb +/mice after 60 muscle stimulations, at the first stimulation. Hindlimb muscle from Chkb -/mice produce less force, and are much more easily fatigued, than that of wild type or Chkb heterozygous mice.
Similar to humans 8,10 , mice with one functional copy of the CHKB gene do not possess any obvious overt muscle dysfunction, whereas mice that are homozygous null for functional copies of the Chkb gene display hallmark muscular dystrophy phenotypes.

Chka protein expression is inversely correlated with the rostro-caudal gradient of severity in Chkb-mediated muscular dystrophy
Consistent with the rostral-to-caudal nature of Chkb associated muscular dystrophy, transmission electron micrographs of 115 day old Chkb -/mice show extensive injury in hindlimb (quadriceps and gastrocnemius) but not the forelimb (triceps) (Supplementary 8 Fig. 1a-c). Chkb encodes choline kinase b, the first enzymatic step in the synthesis of PC, the most abundant phospholipid present in eukaryotic membranes. A second choline kinase, Chka is present in mouse (and human) tissues. We investigated whether the lack of dystrophic phenotypes in Chkb +/mice, and the rostro-caudal gradient of muscular dystrophy in Chkb -/muscle, can be explained by compensatory changes in Chkb or Chka protein levels using western blot. In Chkb +/mice, there was a ~50% decrease in Chkb protein detected in both the forelimb and hindlimb muscles of Chkb +/mice compared to wild type ( Fig. 3a, b). There was no change in Chka protein level in hindlimb muscle of Chkb +/mice compared to wild type, and a small but statistically insignificant increase in Chka level in forelimb muscle.
In Chkb -/mouse forelimb or hindlimb muscle, Chkb protein expression was undetectable consistent with the allele not producing Chkb protein. In forelimb muscle from Chkb -/mice there was a compensatory upregulation of Chka protein expression to almost 3-fold that observed in wild type mice. In contrast, in hindlimb muscle from Chkb -/mice Chka protein expression was decreased to less than 10% that observed in wild type mice. A compensatory level of Chka protein expression inversely correlates with the rostro-caudal gradient of severity in Chkb -/associated muscular dystrophy.

Loss of Chkb activity exerts a major effect on neutral lipid abundance
PC synthesis is integrated with the synthesis of other major phospholipid classes, as well as AcCa, fatty acids and the neutral lipids diacylglycerol and triacylglycerol (Fig.1).
Lipidomics was used to determine if complete loss of Chkb function, and the associated upregulation of Chka in the forelimb but not hindlimb muscle of Chkb -/mice, differentially altered lipid metabolism. The levels of the major glycerophospholipids, neutral lipids and acylcarnitine in hindlimb and forelimb muscle isolated from 12 day old and 30 day old Chkb +/+ and Chkb -/mice were quantified.
In the forelimb and hindlimb muscle of both 12 day old and 30 day old Chkb -/mice, the level of PC was the same as wild type mice (Fig. 4a-d). In 12 day old Chkb -/mice the largest change observed was a 15-fold increase in AcCa level in hindlimb muscle, and to a lesser degree (~2-fold increase) in forelimb, compared to their wild type littermates. The second largest change in 12 day old mice was a 10-fold increase in the level of cardiolipin (CL) in hindlimb muscle that was not present in forelimb muscle of Chkb -/mice. Phosphatidylethanolamine (PE) and phosphatidylinositol (PI) levels were also slightly increased (~1.5 fold) in both forelimb and hindlimb muscles of 12 day old Chkb -/mice. The large changes in lipid levels in hindlimb muscle, versus forelimb, of Chkb -/mice are consistent with the rostral-to-caudal nature of the muscular dystrophy observed in these mice.
Considering the progressive nature of the disease, we tracked the changes in the lipid profile in the hindlimb of 30 day old Chkb -/mice, when muscle injury is more pronounced. In sharp contrast to 12 day old mice, AcCa and CL levels were no longer increased and were at the same level as wild type mice. Instead, there was a 12-fold increase in the neutral storage lipid TG and a 3-fold increase in its precursor DG in the hindlimb samples of Chkb -/mice (Fig. 4e, f). PE and PS levels were 2-3-fold higher in the hindlimb samples from 30 day old Chkb -/mice compared to wild type littermates.
There is a temporal shift from a 12 to 15 fold increase in CL and AcCa, to a similar increase in TG, only in affected muscle in Chkb -/mice.
As AcCa levels are many fold higher than wild type mice in the early stage of Chkb -/muscular dystrophy, this implies that the affected muscles are defective in using fatty acids for the production of cellular energy by mitochondrial β-oxidation. As Chkb -/muscular dystrophy progresses, the affected muscles appear to adapt to this inability to consume fatty acids by transitioning toward energy storage indicated by the large increase in TG.

Increased intramyocellular lipid droplet accumulation and enlarged mitochondria in hindlimb muscles from Chkb -/mice
To understand early ultrastructural pathological changes, and to further explore the associated with mitochondria ( Fig. 5a and Supplementary Fig. 2c). In 115 day old Chkb -/mice, cytoplasmic lipid droplets increased substantially in size (Fig. 5a).
We also evaluated TG accumulation in muscle using confocal microscopy by staining hindlimb muscle sections of 30 day old Chkb -/mice with BODIPY 493/503 (Fig.   5b). Concanavalin A dye conjugate (CF™ 633) and Dapi were used to stain membrane (Red) and nucleus (Blue) respectively. Consistent with our TEM and lipidomics results, BODIPY-stained lipid droplets were noticeably more frequent and larger in Chkb -/hindlimb muscles compared to the wild type littermates. The same pattern of lipid droplet staining was observed using Nile red staining ( Supplementary Fig. 2a, b).
The lipidomics results point to large changes in mitochondrial specific lipids at the early stages of Chkb associated muscular dystrophy. To further explore the nature of these changes, we investigated the temporal development of morphological changes in mitochondria in hindlimb muscle of Chkb -/mice using standard TEM stereological methods 25 . The results show that at 12 days of age, the size of mitochondria increased

Chkb deficiency results in increased lipid droplet accumulation in differentiated myocytes in culture
To address if the observed increase in TG in Chkb -/mice was due to muscle specific events or was due to larger physiological changes that then impact muscle physiology, we assessed TG level in primary cultured muscle cells subsequent to myoblast differentiation.
We first determined if Chkb deficiency alters differentiation in primary myoblasts.
Primary muscle cell cultures were examined for their transition from a single cell proliferative condition to differentiated multinucleated myotubes. During the process of differentiation, mononuclear myoblasts fuse to form myocytes (myotubes), which are large multinucleated cells. We isolated skeletal myoblasts from Chkb +/+ and Chkb -/mice and induced differentiation by switching to low growth factor serum. Representative light micrographs of cultures of dissociated myogenic cells from skeletal muscle of Chkb +/+ and Chkb -/mice at 0, 3 and 5 days after switching to differentiation media show a similar degree of myotube formation (Fig. 6a). Chkb deficiency resulted in a compensatory upregulation of Chka gene expression as well as a significant increase in the markers of myocyte injury, namely Icam1 and Tgfb1 26 (Fig. 6b). We calculated the fusion index, which is nuclei distribution, to determine the extent of myotube differentiation, by immunofluorescence staining. There was no difference between the Ppara and Pparb/d primarily regulate the expression of genes required for fatty acid oxidation, with Pparb/d also regulating genes required for mitochondria biogenesis.
Pparg is primarily expressed in adipose tissue and regulates insulin sensitivity and glucose metabolism 27 . Using reverse transcription (RT) qPCR, we determined that the expression of Ppara and Pparb/d were 4-fold and 6-fold lower, while Pparg was 2-fold higher, in the hindlimb muscle of 30 day old mice Chkb -/mice compared to wild type Ppara and Ppar b/d are the major transcriptional reporters that regulate expression of fatty acid metabolizing genes. The many-fold decrease in the expression of these Ppars that was specific to affected muscle, along with their coreceptors and downstream target genes corroborate the lipdomics data that suggest that the major change in lipid metabolism in Chkb mediated muscular dystrophy is an inability to metabolize fatty acids via mitochondrial β-oxidation resulting in shunting of excess fatty acid into TG rich lipid droplets.

Discussion
Lipid metabolism is highly integrated. Fluctuating levels of lipid metabolites can not only alter shunting of lipids between tangential pathways, but lipids can also directly bind to transcription factors and alter gene expression of lipid metabolic genes. This study highlights these metabolic factors by determining that inactivation of a gene for PC synthesis does not alter PC level. Indeed, the changes in the level of PC do not appear to contribute to the disease phenotype. This study proposes (1) that a change in PC level is not the major metabolic driver behind this disease despite the fact that the genetic defect lies within the major metabolic pathway for the synthesis of PC, (2) a mechanistic model for defective muscle lipid metabolism in Chkb -/mice in which the balance between storage and usage of fatty acids is disrupted, and (3) a mechanism for the rostral-to-caudal gradient for Chkb mediated muscular dystrophy.
Importantly, we report that at an early stage of Chkb mediated muscular dystrophy, there is a 12-to 15-fold increase in the levels of the mitochondrial specific lipids CL and AcCa. Importantly, these changes were observed only in affected muscle of Chkb -/mice. As the disease progresses, AcCa and CL levels return to wild type, and a 12-fold increase in the storage lipid TG occurs. The increase in the mitochondrial specific phospholipid CL is quite telling as far as disease progression. Our TEM of mitochondria in affected muscle during the early stage of Chkb mediated muscular dystrophy revealed a similar number of mitochondria with intact cristae in compared to wild type, however, there was a substantive increase in large mitochondria in affected muscle of Chkb -/mice. We propose that the large increase in CL in affected muscle of Chkb -/mice in the early stage of the disease are mainly driven by the increase in mitochondrial size. As the mice aged the level of CL decreased and had returned to that of wild type by 30 days. At 30 days, mitochondrial size was still increased, however, the number of mitochondria, as well as their cristate (where the bulk of CL resides) were substantively decreased, providing a reasonable explanation for CL mass being reduced to wild type level as the Chkb -/mice aged. Previous observations of mitochondria in Chkb -/mice have only been determined in mice with advanced disease 6,30 , where similar changes in mitochondrial morphological features and numbers were observed. Indeed, one would predict that as Chkb -/mice aged there would be a further decrease in CL mass as mitochondrial numbers further decrease.
Beyond the large increase in CL mass, the other major change in lipid level at the early stage of Chkb mediated muscular dystrophy was a 15-fold increase in AcCa level in affected muscle. This implies that there is either a decreased ability to transport of AcCa into mitochondria for subsequent fatty acid β-oxidation, and/or incomplete βoxidation resulting in a backup of substrate within this pathway. In support of this idea the expression of many of the enzymes required for fatty acid transport into mitochondria and subsequent fatty acid b-oxidation were decreased many fold in affected muscle of Chkb -/mice. The increase in AcCa level at the early stage of Chkb mediated muscular dystrophy, and the decreased expression of genes required for its synthesis and use, is consistent with an inability to import AcCa into mitochondria for fatty acid b-oxidation.
As Chkb -/mice aged, AcCa and CL and levels in hindlimb muscle returned to wild type and by 30 days a dramatic 12-fold increase in TG level was observed. The increase in TG level is consistent with impaired AcCa uptake into mitochondria resulting in a shunting of fatty acids from energy source to energy storage 31 . This observation is consistent with other reports showing that inhibition of PC biosynthesis in mouse liver, and cell culture, significantly increased TG level 32,33 . One interesting additional observation from our study was that ~ 80% of the photographed lipid droplets from Chkb -/hindlimb muscles were closely associated with mitochondria ( Fig. 5a and Supplementary Fig. 2c)

Acknowledgments
We acknowledge funding support from the Canadian Institutes for Health Research (to CRM) and the Atlantic Innovation Fund (to CRM and EH). We thank Gregory Cox for sharing Chkb mice.

DECLARATION OF INTERESTS
The authors declare no competing interests.

In vivo grip strength and fatigability measurements.
Forelimb grip strength was measured using a grip strength meter (Columbus Instruments, Columbus, OH, USA) at 3 time points (6, 12, 18 weeks old) as previously described 40 . All mice were acclimated for a period of five consecutive days before testing. For each time point, Force measurements were collected in the morning hours over a 5-day period, with maximum values for each day over this period averaged to obtain absolute GSM values (Kgf) or normalized to BW (recorded on the first day of testing) for normalized GSM values (Kgf/kg). For the treadmill exhaustion assay, mice are subjected to an enforced running paradigm that tests the resistance level of fatigue in mice. The exhaustion test was performed at 3 time points (7, 13, 19 weeks old) in each group. Groups of mice were made to run on a horizontal treadmill for 5 min at 5 m/min, followed by an increase in the speed of 1m/min each minute. The total distance run by each mouse until exhaustion was measured. Exhaustion was defined as the inability of the mouse to continue running on the treadmill for 30 seconds, despite repeated gentle stimulation.

Primary myoblast isolation, culture and differentiation.
We followed a protocol outlined in Shahini et al. 41 for isolation of myoblast by enabling the outgrowth of these cells from muscle tissue fragments of Chkb +/+ and Chkb -/mice.
Briefly, the mice were euthanized via CO2, were sprayed with 70% ethanol and transferred to a sterile hood. The forelimb and hindlimb muscles were removed, finely minced into small pieces and transferred to a 50 ml conical tube. 1 ml enzymatic solution of PBS containing collagenase type II (500 U/mL), collagenase D (1.5 U/mL), dispase II (2.5 U/mL), and CaCl2 (2.5 mM) was added to the tube. The muscle mixture was placed in a water bath at 37°C for 60 minutes with agitation every 5 minutes. The suspension was centrifuged for 10 minutes at 300 g. Following centrifugation, the supernatant was removed and discarded, and the pellet was resuspended in to allow attachment of the tissues to the surface and subsequent outgrowth and migration of cells. The myogenic cell population was further purified with one round of pre-plating on collagen coated dishes to isolate fibroblasts from myoblasts. To induce differentiation into multinucleated myotubes, the cells were seeded at 10000 cells/cm 2 on plastic coverslip chambers coated with Matrigel and the medium was replaced by differentiation medium containing DMEM with high glucose and 5% HS.

Ex vivo force measurement.
At the end of the in vivo phase (Week 19), mice were deeply anesthetized with ketamine and xylazine (80 and 10 mg/kg). The extensor digitorum longus (EDL) muscle of the right hindlimb was removed for comparison of Ex vivo force contractions between groups as previously described 42,43 . Briefly, the EDL muscle was securely tied with braided surgical silk at both tendon insertions to the lever arm of a servomotor/force transducer (model 305B) (Aurora Scientific, Aurora, Ontario, Canada) and the proximal tendon was fixed to a stationary post in a bath containing buffered Ringer solution (composition in mM: 137 NaCl, 24 NaHCO3, 11 glucose, 5 KCl, 2 CaCl2, 1 MgSO4, 1 NaH2PO4 and 0.025 turbocurarine chloride) maintained at 25˚C and bubbled with 95% O2 -5% CO2 to stabilize pH at 7.4. At optimal muscle length, the maximal force developed was measured during trains of stimulation (300 milliseconds, ms) with increasing frequencies up to 250 Hz or until the highest plateau was achieved. The force generated to obtain the highest plateau was used to determine specific force (maximal force normalized to cross-sectional area of the muscle). Finally, the muscle was subjected to a fatigue protocol consisting of 60 isometric contractions for 300 ms each, once every 5 seconds. The frequency at which the EDL muscles were stimulated is 250 Hz. The force was recorded every 10th contraction during the repetitive contractions and again at 5 and 10 min afterward to measure recovery. Samples were assigned to controls and test groups. CT values were normalized based on a/an Manual Selection of reference genes. The data analysis web portal calculates fold change/regulation using delta delta CT method, in which delta CT is calculated between gene of interest (GOI) and an average of reference genes (HKG), followed by delta-delta CT calculations (delta CT (Test Group)-delta CT (Control Group)). Fold Change is then calculated using 2^ (-delta delta CT) formula. The data analysis web portal to plot scatter clustergram, and heat map.

Lipid extraction
We performed lipid extractions using the modified Bligh and Dyer extraction for LC-MS analysis of lipids protocol 44 . All reagents were of LC-MS grade. Briefly, the muscle tissue (~10mg) was homogenized with a steel bead in 1 ml of cold 0.1 N HCl:methanol 70:30%) was injected onto the column. The following system gradient was used for separating the lipid classes and molecular species: 30% solvent B for 3 min; then solvent B increased to 50% over 6 min, then to 70% B in 6 min, then kept at 99% B for 20 min, and finally the column was re-equilibrated to starting conditions (30% solvent A) for 5 min prior to each new injection.

High resolution tandem mass spectrometry and lipidomics.
Lipid analyses were carried out using a Q-Exactive Orbitrap mass spectrometer controlled by X-Calibur software 4.0 (ThermoScientific, MO, USA) with an acquisition HPLC system. The following parameters were used for the Q-Exactive mass spectrometer -sheath gas: 40, auxiliary gas: 5, ion spray voltage: 3. First, the individual data files were searched for product ion MS/MS spectra of lipid precursor ions. MS/MS fragment ions were predicted for all precursor adduct ions measured within ±5 ppm. The product ions that matched the predicted fragment ions within a ±5 ppm mass tolerance was used to calculate a match-score, and those candidates providing the highest quality match were determined. Next, the search results from the individual positive or negative ion files from each sample group were aligned within a retention time window (±0.2 min) and the data were merged for each annotated lipid.

Data cleanup and statistical analysis of lipids.
Lipid concentrations extracted from the LipidSearch software were further analyzed with an in-house script using the R programming language. The data was filtered to exclude any peak concentration estimates with a signal to noise ratio (SNR parameter) of less than 2.0 or a peak quality score (PQ parameter) of less than 0.8. If this exclusion resulted in the removal of two observation within a biological triplicate, the remaining observation was also excluded. The individual concentrations were then gathered together by lipid identity (summing together the concentration of multiple mass spectrometry adducts where these adducts originated from the same molecular source and averaging together biological replicates) and grouped within the broader categories of AcCa, TG, DG, PC, PE, PG, CL, PI, PS. The result was nine groups containing multiple lipid concentrations corresponding to specific lipid identities, which were then compared between wild type and KO samples using a (paired, non-parametric) Wilcoxon signed-rank test at an overall significance level of 5% (using the Bonferroni correction to account for the large number of tests performed). As the Bonferroni correction is fairly conservative, significant differences are reported at both precorrection (*) and post-correction (***) significance levels. Microscope at 80kV. Images were captured using a Hamamatsu ORCA-HR digital camera. Three mice per genotype for each timepoint were evaluated. The mitochondrial content was determined from the images at 10,000× magnification using Image J software and calculated as mitochondria count/field by blinded investigators. Point counting was used to estimate mitochondrial volume density and mitochondrial cristae density based on standard stereological methods 46,47 . Only mitochondria profiles of acceptable quality defined as clear visibility and no or few missing spots of the inner membrane were included. Using ImageJ software, a point grid was digitally layered over the micrographic images at 20,000x or 40,000x magnification for mitochondrial volume density and cristae density calculations respectively. Grid sizes of 85 nm x 85 nm and 165 nm x 165 nm were used to estimate mitochondria volume and cristae surface area, respectively. Mitochondria volume density was calculated by dividing the points assigned to mitochondria to the total number of points counted inside the muscle. The mitochondrial cristae surface area per mitochondrial volume (mitochondrial cristae density) was estimated by the formula: mitochondrial cristae density = (4/π) BA, where BA is the boundary length density estimated by counting intersections on test lines multiplied by π/2. In brief, we counted the intersections I(imi) between the inner mitochondrial membrane trace and the test lines and measured the total length of the test line within the mitochondria profile to calculate mitochondrial cristae density =2. I(imi)/L(mi).

Western blot analysis (WB) and quantification.
The muscle tissue (~100mg) was homogenized with a steel bead in 1 ml of cold RIPA

Quantification and statistical analysis.
All experiments were repeated 3 or more times. Data are presented as mean ± SEM or mean ± SD, as appropriate. For comparison of two groups the two-tailed Student's t-test was used unless otherwise specified. Comparison of more than two groups was done by one-way ANOVA followed by the Tukey's Multiple Comparison test. P values <0.05 were considered significant.

Data availability
All data that support the findings of this study are available from the corresponding authors upon request.